Data

Tools

Indexes

Applications

Experiments

Open Source Science

diffusion_ai

This small library covers the Stable Diffusion algorithm and its variants.

https://github.com/khethan123/diffusion_ai

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (11.2%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

This small library covers the Stable Diffusion algorithm and its variants.

Basic Info
  • Host: GitHub
  • Owner: khethan123
  • License: apache-2.0
  • Language: Jupyter Notebook
  • Default Branch: main
  • Size: 77.1 KB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created over 1 year ago · Last pushed over 1 year ago
Metadata Files
Readme License Citation

README.md

Diffusion AI

Overview

Welcome to diffusion_ai! This small library is the result of my learning journey from the Practical Deep Learning for Coders course. This course, created by Jeremy Howard, Jonathan Whitaker, and Tanishq Abraham, covers the Stable Diffusion algorithm and its variants from scratch. Using this course, I deepened my understanding of diffusion models and even implemented them on my own. You can do the same by using this library.

Predicted Images and Videos

Original image
original_dataset

Here are some predicted images and videos generated using `diffusion_ai`:
Uncoditional Generation
predicted_image


Class Conditioned generation
predicted_class


DDPM denoising process

Getting Started

To run diffusion_ai, follow these steps:

  1. Clone the Repository: bash git clone https://github.com/khethan123/diffusion_ai.git cd diffusion_ai

  2. Install Requirements: bash pip install -r requirements.txt

  3. Follow the Jupyter Notebook Tutorials: Explore the tutorials to implement DDPM, DDIM, Unet, timestep embeddings, class-conditioned outputs, and training with VAE latents.

Example Results

Performance on Fashion MNIST Dataset

  • FID Score: 4.058 on generated data
  • Model: Unet with timestep embeddings, attention block, and class conditioning
  • Training Loss: 3.2%
  • Evaluation Loss: 3.3%
  • Training Duration: 25 epochs
  • Sampler: DDIM
  • Image Size: 32x32 pixels
  • Time Steps: 100

Get Started Now!

Dive into the exciting world of diffusion models and deep learning by getting started with diffusion_ai today. Happy coding!

Acknowledgements

This project was made possible by the "Practical Deep Learning for Coders" course by Jeremy Howard, Jonathan Whitaker, and Tanishq Abraham, PhD. Their teachings provided the foundation for this library.

Citations

This project was made possible by the "Practical Deep Learning for Coders" course by course.fast.ai If you use this library or any part of the course content, please cite:

Citation: Jeremy Howard, Jonathan Whitaker, and Tanishq Abraham. "Practical Deep Learning for Coders." course.fast.ai, 2022, part 2 Accessed 21 June 2024. See CITATION file for more details.

License

This project incorporates content licensed under the Apache 2.0 License from the "Practical Deep Learning for Coders" course. The full text of the Aphache 2.0 License is present in the repository.

For any questions or contributions, please feel free to open an issue or pull request.

Owner

  • Login: khethan123
  • Kind: user

Citation (CITATION.cff) 

cff-version: 1.2.0
message: If you use this library or the any part of the course content, please cite it using these metadata.
title: Practical Deep Learning for Coders
abstract: This course provides comprehensive coverage of deep learning techniques, including Stable Diffusion and its variants, along with practical implementation in Python and PyTorch.
authors:
  - family-names: Howard
    given-names: Jeremy
  - family-names: Whitaker
    given-names: Jonathan
  - family-names: Abraham
    given-names: Tanishq
date-released: "2024-06-12"  # this is the date when I completed the course
identifiers:
  - description: Practical Deep Learning for Coders course materials
    type: url
    value: "https://course.fast.ai/Lessons/part2.html"
license: Apache-2.0
repository-code: "https://github.com/fastai/course22p2"

GitHub Events

Total
Last Year

Dependencies

requirements.txt pypi
  • accelerate *
  • datasets *
  • diffusers *
  • einops *
  • fastai *
  • fastcore *
  • ffmpeg-python *
  • matplotlib *
  • numpy *
  • pillow *
  • scipy *
  • timm *
  • torch *
  • torcheval *
  • tqdm *